How often do pages rating within the SERPs get cited in AI Overviews?
We analyzed 1.9M citations final yr in an try and reply that query.
However, as with every part AI-related, so much has modified since then.
For example, as of January 2026, AI Overviews at the moment are powered by Gemini 3 to higher reply searchers’ long-tail questions—having beforehand relied on earlier Gemini fashions, together with 2.5 variants.
Since that final research we’ve additionally improved our parsing methodology in Ahrefs, in order that we are able to see much more of the citations that seem in AI Overviews.
So, we determined it was time to replace our findings to higher reply that preliminary query.
This time round, we analyzed 863K key phrase SERPs, and a grand complete of 4M AI Overview URLs—over double our final evaluation.
All of this knowledge was taken from Ahrefs Model Radar, our AI visibility device.
Right here’s what we discovered…
In our evaluation, we checked out instances the place the identical URL appeared in each the AI Overview and the common SERP, for the identical question.
In our first check, we studied all returning outcomes on the web page; together with common blue hyperlinks and SERP options.
So, for instance, we tracked Adverts, Featured snippets, PAA packing containers, Video Packs and Natural Listings as separate blocks.
From this, we found that 37.9% of URLs cited in AI Overviews additionally appeared throughout the first 10 blocks.
The remaining have been nearly evenly break up between positions 11–100 (31.2%) and past the highest 100 blocks (31.0%).
In our second check, we centered on customary blue hyperlinks solely and ignored Adverts and different SERP options. The distribution ended up being largely related:
| Natural place | % of cited URLs |
|---|---|
| Rank in high 10 | 37.10% |
| Rank 11–100 | 26.20% |
| Don’t rank in high 100 | 36.70% |
The one slight distinction was that extra AI Overview–cited pages fell exterior of the highest 100.
This means that AIOs generally cite pages that don’t seem in common search outcomes, however do seem in SERP Options.
So, what does all of this knowledge really imply?
Effectively, not less than compared to our preliminary research, it signifies that Google is deciding on far fewer pages straight from the unique SERP (~76% in July 2025 vs. ~38% at the moment).
To me, that means AI Overviews are relying much less on the direct search outcomes and extra on the sources exhibiting up in fan out question SERPs.
Google has confirmed that its system performs a “Question fan-out” each time a consumer searches and AI is triggered.
That is when the preliminary question is break up into a number of associated sub-queries.
The pages that seem most frequently inside these sub-query SERPs then get cited within the AI Overview.
So, with the arrival of Gemini 3, it could be that fan out queries are taking part in an even bigger function in supply choice.
It could possibly be increasing queries extra aggressively than earlier than, or increasing them extra broadly—pulling from associated SERPs the place fewer pages rank for the unique question.
Our analysis additionally revealed one thing fairly attention-grabbing about YouTube.
Among the many AI Overview cited pages that didn’t rank in Google’s high 100 outcomes for a similar key phrase, 18.2% have been YouTube URLs.
And people YouTube URLs accounted for five.6% of all AI Overview URLs cited throughout the dataset.
In different phrases, YouTube makes up a significant slice of complete AI Overview citations, regardless of not rating within the SERPs for the direct question.
In response to knowledge in Ahrefs Model Radar, YouTube is essentially the most cited area in AI Overviews at the moment, and has grown 34% during the last six months.
A few of our different research reinforce the significance of YouTube for AI visibility.
For instance, our analysis of 75K manufacturers revealed that mentions on YouTube—in video titles, transcripts, and descriptions—are the strongest correlating issue with AI Overview visibility.
Bearing all of this in thoughts, you’d be sensible to make YouTube a part of your AI search technique.
You’ll be able to see which precise YouTube movies point out your model in Model Radar.
And, you may even see which of these movies get drawn into AI Overviews by checking the “Cited in AI” column.
If rating in the identical SERPs as an AI Overview is not sufficient to win an AIO quotation, then what is?
I just lately wrote an article on find out how to rank in AI Overviews.
In it, I spoke about how SEOs are shifting from optimizing content material for key phrases, to optimizing for fan-out queries.
search engine optimization Marketing consultant, Ethan Lazuk, writes about precisely this in his nice article Google’s Question Fan-Out Approach and What SEOs Ought to Know About It:
The extra related you can also make the passages in your paperwork (or different media) to fan-out queries, the likelier you’ll be to earn a point out or a quotation within the AI-generated reply. So for the query of how question fan-out impacts rankings, simply know that we’re not optimizing for particular person key phrases however reasonably whole consumer journeys, and people fan-out queries information the manner.
The factor is, Google doesn’t expose the simulated fan-out queries they generate for every AI response.
And, even when they did, AI Overviews are probabilistic.
Their content material, citations, entities, and fan-outs change with each question. Learn our research AI Overviews Change Each 2 Days (However By no means Change Their Thoughts) for extra on that.
However there are methods you may replicate the question fan-out course of for your self utilizing instruments, together with:
- Qforia from iPullRank Founder Mike King
- This Gemini API + Screaming Frog workflow from GoFishDigital Co-Founder Dan Hinkley
- AI Visibility Fan Out by WordLift
No matter route you select, it’s essential be sure to’re protecting your subject from all angles.
Listed here are some extra methods you are able to do that utilizing Ahrefs:
Use Guardian Subjects to cowl each base
Guardian Subjects in Ahrefs Key phrases Explorer reveals you all of the key phrases, questions, and intent angles you would doubtlessly goal with a single piece of content material.
Use it to seek out and cluster your content material round associated themes—not simply single key phrases.
Examine fan-out queries generated by different AI instruments
Fan-out queries are simulated and alter from run to run.
It doesn’t matter a lot whether or not you handle to completely mirror Gemini’s inside expansions.
It issues extra that you’ll be able to analyze a number of enlargement examples, at scale, to establish recurring themes, subjects, and intent angles that an AI could contemplate when setting up its reply.
You’ll be able to research tens of millions of ChatGPT and Perplexity fan-out question examples—tied to each database prompts and your personal customized prompts—throughout the AI Responses report in Model Radar.
Use AI Content material Helper to create complete content material
Our staff has designed an AI Content material Helper that allows you to optimize your content material particularly for AI search.
It generates associated fan-out queries, then measures the cosine similarity between the subjects lined in your content material, and the subjects the SERP or AI response is probably going attempting to deal with.
As you write, a coloured spotlight seems exhibiting you ways totally you cowl every subject.
You should use it to benchmark your content material towards the highest 10 SERP outcomes—however, as this very analysis has made clear, usually the pages you really compete with aren’t all the time rating on web page one; generally they rank additional down, generally they seem in wealthy SERP options, and generally they’re picked from a special SERP altogether.
For that motive, now you can benchmark your content material towards your personal customized rivals in Ahrefs’ AI Content material Helper—together with people who often win AI Overview citations.
Assessment your AI Overview citations in Model Radar
Learning your personal AI Overview citations can assist you declare extra visibility as a result of it reveals you precisely what the system already trusts you for—and the place you’re shut however not fairly chosen.
For instance, you can:
- See which pages and subjects are already being cited, which reveals the angles Google considers you authoritative on.
- Establish gaps the place rivals are cited as an alternative, and reverse-engineer what extra context, entities, or subtopics they’re protecting.
- Spot patterns in format (definitions, stats, step-by-steps, video embeds) that are likely to get pulled into AI solutions.
- Strengthen pages which can be rating however not being cited—usually an indication that topical protection or readability doesn’t absolutely align with the expanded intent.
Briefly, quotation evaluation strikes you from “rating properly” to “being chosen as a supply,” which as we all know, could not all the time be the identical factor.
Ultimate ideas
AI Overviews have shifted, and rating for the consumer’s precise question is not a assure of visibility.
If you wish to perceive what Google really trusts and find out how to earn extra citations, begin by learning the patterns in your personal knowledge.
And, as standard, in case you have any questions on this analysis, be happy to ping me on LinkedIn.
